Cross section recondensation method via generalized energy condensation theory

The standard multigroup method used in whole-core reactor analysis relies on energy condensed (coarse-group) cross sections generated from single lattice cell calculations, typically with specular reflective boundary conditions. Because these boundary conditions are an approximation and not representative of the core environment for that lattice, an error is introduced in the core solution (both eigenvalue and flux). As current and next generation reactors trend toward increasing assembly and core heterogeneity, this error becomes more significant. The method presented here corrects for this error by generating updated coarse-group cross sections on-the-fly within whole-core reactor calculations without resorting to additional cell calculations. In this paper, the fine-group core flux is unfolded by making use of the recently published Generalized Condensation Theory and the cross sections are recondensed at the whole-core level. By iteratively performing this recondensation, an improved core solution is found in which the core-environment has been fully taken into account. This recondensation method is both easy to implement and computationally very efficient because it requires precomputation and storage of only the energy integrals and fine-group cross sections. In this work, the theoretical basis and development of cross section recondensation is presented, and the method is verified with several sample problems.

► A new method is presented which corrects for core environment error from specular boundaries at the lattice cell level.
► Solution obtained with generalized energy condensation provides improved approximation to the core level fine-group flux.
► Iterative recondensation of the cross sections and unfolding of the flux provides on-the-fly updating of the core cross sections.
► Precomputation of energy integrals and fine-group cross sections allows for easy implementation and efficient solution.
► Method has been implemented in 1D and shown to correct the environment error, particularly in strongly heterogeneous cores.